Merge pull request #186 from smith-chem-wisc/StefanBranch

fixes

Author: stefanks

InternalLogic/AnalysisEngine.cs

@@ -16,7 +16,6 @@ public class AnalysisEngine : MyEngine
 
         #region Private Fields
 
-        private const double comboThresholdMultiplier = 3;
         private const int max_mods_for_peptide = 3;
         private readonly double binTol;
         private readonly int maximumMissedCleavages;
@@ -299,7 +298,6 @@ public void ApplyProteinParsimony(out List<ProteinGroup> proteinGroups)
                 proteinListHere.Add(kvp.Key);
                 proteinGroups.Add(new ProteinGroup(proteinListHere, kvp.Value, uniquePeptidesHere));
             }
-            
 
             // grab indistinguishable proteins
             foreach (var proteinGroup in proteinGroups)
@@ -396,15 +394,15 @@ public void ScoreProteinGroups(List<ProteinGroup> proteinGroups, List<NewPsmWith
             }
 
             // add every psm that corresponds to the protein group's peptides to the group
-            foreach(var proteinGroup in proteinGroups)
+            foreach (var proteinGroup in proteinGroups)
             {
-                foreach(var peptide in proteinGroup.PeptideList)
+                foreach (var peptide in proteinGroup.PeptideList)
                 {
                     string peptideBaseSequence = string.Join("", peptide.BaseSequence.Select(b => char.ConvertFromUtf32(b)));
                     List<NewPsmWithFdr> psmListForThisBaseSeq = new List<NewPsmWithFdr>();
 
                     peptideBaseSeqToPsmMatching.TryGetValue(peptideBaseSequence, out psmListForThisBaseSeq);
-                    foreach(var psm in psmListForThisBaseSeq)
+                    foreach (var psm in psmListForThisBaseSeq)
                     {
                         if (!proteinGroup.TotalPsmList.Contains(psm))
                         {
@@ -488,24 +486,24 @@ public void ScoreProteinGroups(List<ProteinGroup> proteinGroups, List<NewPsmWith
 
                 // score the group (scoring algorithm defined in the ProteinGroup class)
                 proteinGroup.ScoreThisProteinGroup();
-                
+
                 // remove empty protein groups (peptides were too poor quality and group doesn't exist anymore)
                 if (proteinGroup.proteinGroupScore == 0)
                     proteinGroupsToRemove.Add(proteinGroup);
             }
 
-            foreach(var proteinGroup in proteinGroupsToRemove)
+            foreach (var proteinGroup in proteinGroupsToRemove)
             {
                 proteinGroups.Remove(proteinGroup);
             }
 
             // build razor peptide list (peptides that have >1 protein groups in the final protein group list)
-            foreach(var kvp in peptideToProteinGroupMatching)
+            foreach (var kvp in peptideToProteinGroupMatching)
             {
                 if (kvp.Value.Count > 1)
                     allRazorPeptides.Add(kvp.Key);
             }
-            
+
             foreach (var proteinGroup in proteinGroups)
             {
                 foreach (var peptide in proteinGroup.PeptideList)
@@ -519,7 +517,7 @@ public void ScoreProteinGroups(List<ProteinGroup> proteinGroups, List<NewPsmWith
                     // build PeptideWithSetMod list to calc sequence coverage
                     HashSet<PeptideWithSetModifications> peptidesWithSetMods = null;
                     compactPeptideToProteinPeptideMatching.TryGetValue(peptide, out peptidesWithSetMods);
-                    foreach(var pep in peptidesWithSetMods)
+                    foreach (var pep in peptidesWithSetMods)
                     {
                         proteinGroup.PeptideWithSetModsList.Add(pep);
                     }
@@ -634,6 +632,28 @@ protected override MyResults RunSpecific()
 
         #region Private Methods
 
+        private static void OverlappingIonSequences(BinTreeStructure myTreeStructure)
+        {
+            foreach (Bin bin in myTreeStructure.FinalBins)
+            {
+                foreach (var hm in bin.uniquePSMs.Where(b => !b.Value.Item3.IsDecoy))
+                {
+                    var ya = hm.Value.Item3.thisPSM.newPsm.matchedIonsList;
+                    if (ya.ContainsKey(ProductType.B) && ya.ContainsKey(ProductType.Y) && ya[ProductType.B].Any(b => b > 0) && ya[ProductType.Y].Any(b => b > 0))
+                        if (ya[ProductType.B].Last(b => b > 0) + ya[ProductType.Y].Last(b => b > 0) > hm.Value.Item3.thisPSM.PeptideMonoisotopicMass)
+                            bin.Overlapping++;
+                }
+            }
+        }
+
+        private static void IdentifyPsmsWithMaxMods(BinTreeStructure myTreeStructure)
+        {
+            foreach (Bin bin in myTreeStructure.FinalBins)
+            {
+                bin.FracWithMaxMods = ((double)bin.uniquePSMs.Values.Count(b => !b.Item3.IsDecoy && b.Item3.thisPSM.NumVariableMods == max_mods_for_peptide)) / bin.CountTarget;
+            }
+        }
+
         private static void IdentifyAAsInCommon(BinTreeStructure myTreeStructure)
         {
             foreach (Bin bin in myTreeStructure.FinalBins)
@@ -771,7 +791,7 @@ private static void IdentifyCombos(BinTreeStructure myTreeStructure, double v)
             var ok = new HashSet<Tuple<double, double, double>>();
             foreach (var bin in myTreeStructure.FinalBins.Where(b => Math.Abs(b.MassShift) > v))
                 foreach (var bin2 in myTreeStructure.FinalBins.Where(b => Math.Abs(b.MassShift) > v))
-                    if (bin.CountTarget * bin2.CountTarget >= totalTargetCount * comboThresholdMultiplier)
+                    if (bin.CountTarget * bin2.CountTarget >= totalTargetCount)
                         ok.Add(new Tuple<double, double, double>(bin.MassShift, bin2.MassShift, Math.Min(bin.CountTarget, bin2.CountTarget)));
 
             foreach (var bin in myTreeStructure.FinalBins)
@@ -875,6 +895,10 @@ private static BinTreeStructure MyAnalysis(List<NewPsmWithFdr> limitedpsms_with_
 
             IdentifyMine(myTreeStructure, binTol);
 
+            IdentifyPsmsWithMaxMods(myTreeStructure);
+
+            OverlappingIonSequences(myTreeStructure);
+
             return myTreeStructure;
         }
 

InternalLogic/Bin.cs

@@ -6,6 +6,7 @@ namespace InternalLogicEngineLayer
 {
     public class Bin
     {
+
         #region Public Fields
 
         public string UnimodId = "-";
@@ -70,6 +71,8 @@ public int LocalizeableTarget
 
         public string Mine { get; internal set; }
         public Dictionary<char, int> AAsInCommon { get; internal set; }
+        public double FracWithMaxMods { get; internal set; }
+        public int Overlapping { get; internal set; }
 
         #endregion Public Properties
 
@@ -97,5 +100,6 @@ internal void Add(NewPsmWithFdr ok)
         }
 
         #endregion Internal Methods
+
     }
 }

InternalLogic/PSMwithProteinHashSet.cs

@@ -174,7 +174,7 @@ public override string ToString()
 
         #region Internal Methods
 
-        internal static double MatchIons(IMsDataScan<IMzSpectrum<MzPeak>> thisScan, Tolerance product_mass_tolerance_value, double[] sorted_theoretical_product_masses_for_this_peptide, double[] matchedIonsList)
+        internal static double MatchIons(IMsDataScan<IMzSpectrum<MzPeak>> thisScan, Tolerance product_mass_tolerance_value, double[] sorted_theoretical_product_masses_for_this_peptide, double[] matchedIonMassesList)
         {
             var TotalProductsHere = sorted_theoretical_product_masses_for_this_peptide.Length;
             if (TotalProductsHere == 0)
@@ -199,12 +199,12 @@ internal static double MatchIons(IMsDataScan<IMzSpectrum<MzPeak>> thisScan, Tole
                 {
                     MatchingProductsHere++;
                     MatchingIntensityHere += experimental_intensities[i];
-                    matchedIonsList[theoreticalIndex] = nextTheoreticalMass;
+                    matchedIonMassesList[theoreticalIndex] = nextTheoreticalMass;
                 }
                 else if (currentExperimentalMZ < nextTheoreticalMZ)
                     continue;
                 else
-                    matchedIonsList[theoreticalIndex] = -nextTheoreticalMass;
+                    matchedIonMassesList[theoreticalIndex] = -nextTheoreticalMass;
                 i--;
                 // Passed a theoretical! Move counter forward
                 theoreticalIndex++;

InternalLogicWithFileIO/Parent/MyTaskEngine.cs

@@ -368,14 +368,13 @@ protected IEnumerable<Protein> GetProteins(bool onTheFlyDecoys, IDictionary<stri
                     while (true)
                     {
                         string line = fasta.ReadLine();
-                        
 
                         if (line.StartsWith(">"))
                         {
                             // fasta protein only has accession, fullname, sequence (no mods)
                             string[] delimiters = { ">", "|", " OS=" };
                             string[] output = line.Split(delimiters, StringSplitOptions.None);
-                            if(output.Length > 4)
+                            if (output.Length > 4)
                             {
                                 accession = output[2];
                                 name = accession;
@@ -387,11 +386,10 @@ protected IEnumerable<Protein> GetProteins(bool onTheFlyDecoys, IDictionary<stri
                                 full_name = line.Substring(1);
                                 accession = "";
                             }
-                            
+
                             // new protein
                             sequence = "";
                         }
-
                         else
                         {
                             sequence += line.Trim();
@@ -489,7 +487,7 @@ protected void WriteTree(BinTreeStructure myTreeStructure, string output_folder,
             var writtenFile = Path.Combine(output_folder, fileName + ".mytsv");
             using (StreamWriter output = new StreamWriter(writtenFile))
             {
-                output.WriteLine("MassShift\tCount\tCountDecoy\tCountTarget\tCountLocalizeableTarget\tCountNonLocalizeableTarget\tFDR\tArea 0.01t\tArea 0.255\tFracLocalizeableTarget\tMine\tUnimodID\tUnimodFormulas\tAA\tCombos\tModsInCommon\tAAsInCommon\tResidues\tNtermLocFrac\tCtermLocFrac\tUniprot");
+                output.WriteLine("MassShift\tCount\tCountDecoy\tCountTarget\tCountLocalizeableTarget\tCountNonLocalizeableTarget\tFDR\tArea 0.01t\tArea 0.255\tFracLocalizeableTarget\tMine\tUnimodID\tUnimodFormulas\tAA\tCombos\tModsInCommon\tAAsInCommon\tResidues\tNtermLocFrac\tCtermLocFrac\tFracWithMaxMods\tOverlappingFrac\tUniprot");
                 foreach (Bin bin in myTreeStructure.FinalBins.OrderByDescending(b => b.Count))
                 {
                     output.WriteLine(bin.MassShift.ToString("F3", CultureInfo.InvariantCulture)
@@ -512,6 +510,8 @@ protected void WriteTree(BinTreeStructure myTreeStructure, string output_folder,
                         + "\t" + string.Join(",", bin.residueCount.OrderByDescending(b => b.Value).Select(b => b.Key + ":" + b.Value))
                         + "\t" + (bin.LocalizeableTarget == 0 ? double.NaN : (double)bin.NlocCount / bin.LocalizeableTarget).ToString("F3", CultureInfo.InvariantCulture)
                         + "\t" + (bin.LocalizeableTarget == 0 ? double.NaN : (double)bin.ClocCount / bin.LocalizeableTarget).ToString("F3", CultureInfo.InvariantCulture)
+                        + "\t" + (bin.FracWithMaxMods).ToString("F3", CultureInfo.InvariantCulture)
+                        + "\t" + ((double)bin.Overlapping / bin.CountTarget).ToString("F3", CultureInfo.InvariantCulture)
                         + "\t" + bin.uniprotID);
                 }
             }

OldInternalLogic/OldInternalLogic.csproj

@@ -116,10 +116,10 @@
   </ItemGroup>
   <ItemGroup>
     <Content Include="combos.txt">
-      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+      <CopyToOutputDirectory>Always</CopyToOutputDirectory>
     </Content>
     <Content Include="f.txt">
-      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+      <CopyToOutputDirectory>Always</CopyToOutputDirectory>
     </Content>
     <Content Include="glyco.txt">
       <CopyToOutputDirectory>Always</CopyToOutputDirectory>
@@ -128,13 +128,13 @@
       <CopyToOutputDirectory>Always</CopyToOutputDirectory>
     </Content>
     <Content Include="r.txt">
-      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+      <CopyToOutputDirectory>Always</CopyToOutputDirectory>
     </Content>
     <Content Include="s.txt">
-      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+      <CopyToOutputDirectory>Always</CopyToOutputDirectory>
     </Content>
     <Content Include="v.txt">
-      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+      <CopyToOutputDirectory>Always</CopyToOutputDirectory>
     </Content>
   </ItemGroup>
   <Import Project="$(MSBuildToolsPath)\Microsoft.CSharp.targets" />

OldInternalLogic/ProteomeDatabaseReader.cs

@@ -3,7 +3,6 @@
 using System.Collections.Generic;
 using System.IO;
 using System.IO.Compression;
-using System.Text;
 using System.Xml;
 
 namespace OldInternalLogic
@@ -28,6 +27,7 @@ static ProteomeDatabaseReader()
             modificationTypeCodes.Add("Peptide N-terminal.", ModificationType.PeptideNTerminus);
             modificationTypeCodes.Add("Peptide C-terminal.", ModificationType.PeptideCTerminus);
             modificationTypeCodes.Add("Anywhere.", ModificationType.AminoAcidResidue);
+            modificationTypeCodes.Add("Protein core.", ModificationType.AminoAcidResidue);
 
             aminoAcidCodes = new Dictionary<string, char>();
             aminoAcidCodes.Add("Alanine", 'A');
@@ -124,7 +124,7 @@ public static IEnumerable<MorpheusModification> ReadModFile(string v)
                             case "PP":
                                 if (feature_type == "MOD_RES")
                                 {
-                                    modificationTypeCodes.TryGetValue(line.Substring(5), out modification_type);
+                                    modification_type = modificationTypeCodes[line.Substring(5)];
                                 }
                                 break;
 
@@ -156,9 +156,9 @@ public static IEnumerable<MorpheusModification> ReadModFile(string v)
                                     if (Math.Abs(monoisotopic_mass_shift - ye.MonoisotopicMass) > 1e-3)
                                         throw new InvalidDataException("In file" + v + " Modification " + description + " mass formula mismatch");
                                     if (labileOrSticky.Equals("Labile") || labileOrSticky.Equals("Both"))
-                                        yield return new MorpheusModification(description, modification_type, amino_acid_residue,  Path.GetFileNameWithoutExtension(v), database_name, prevAA, alternative_mass, true, ye);
+                                        yield return new MorpheusModification(description, modification_type, amino_acid_residue, Path.GetFileNameWithoutExtension(v), database_name, prevAA, alternative_mass, true, ye);
                                     if (labileOrSticky.Equals("Sticky") || labileOrSticky.Equals("Both"))
-                                        yield return new MorpheusModification(description, modification_type, amino_acid_residue,  Path.GetFileNameWithoutExtension(v), database_name, prevAA, alternative_mass, false, ye);
+                                        yield return new MorpheusModification(description, modification_type, amino_acid_residue, Path.GetFileNameWithoutExtension(v), database_name, prevAA, alternative_mass, false, ye);
                                 }
                                 description = null;
                                 feature_type = null;

OldInternalLogic/combos.txt

@@ -12,4 +12,5 @@
 -17.02655 15.994915
 -17.026549 0.984016
 21.981943 21.981943
-14.015650 0.984016
+14.015650 0.984016
+79.966331 79.966331

OldInternalLogic/m.txt

@@ -74,7 +74,7 @@ CF   H-3 N-1
 ID   Loss of two hydrogens from N term M
 FT   MOD_RES
 TG   Methionine
-PP   Protein N-terminal.
+PP   N-terminal.
 SL   Sticky
 MM   -2.016
 CF   H-2
@@ -469,7 +469,7 @@ CF   C6 H12 N2 O1
 ID   Acetylation+Methionine
 FT   MOD_RES
 TG   Any
-PP   Protein N-terminal.
+PP   N-terminal.
 SL   Sticky
 MM   173.051055
 CF   C7 H11 N1 O2 S
@@ -485,7 +485,7 @@ CF   H24 C15
 ID   Myristoylation
 FT   MOD_RES
 TG   Glycine
-PP   Protein N-terminal.
+PP   N-terminal.
 SL   Both
 MM   210.198366
 CF   H26 C14 O1
@@ -541,7 +541,7 @@ CF   H30 C16 O1
 ID   Palmitoylation
 FT   MOD_RES
 TG   Any
-PP   Protein N-terminal.
+PP   N-terminal.
 SL   Both
 MM   238.229666
 CF   H30 C16 O1